Background. Prevention of central nervous system (CNS) leukemia by early introduction of therapy to this sanctuary site is an essential component of modern treatment strategy for acute lymphoblastic leukemia (ALL). However, the optimal form of preventive CNS therapy remains debatable. Procedure. To address this issue, we evaluated the efficacy of CNS preventive therapy for 572 children with ALL who achieved complete remission in the Children's Cancer and Leukemia Study Group (CCLSG) ALL874 ( -1990( ) and ALL911 (1991( -1993) ) studies. They received risk-directed therapy based on age and leukocyte count. In the ALL 874 study, the non-high-risk (low-risk [LR] + intermediate risk [IR]) patients were randomly assigned to the conventional cranial irradiation (CRT) regimen (L874A and I874A) and the highdose methotrexate (HDMTX) regimen without CRT (L874B and I874B). The former patients received 18-Gy CRT plus 3 doses of intrathecal (IT) MTX and the latter patients received 3 courses of HDMTX at 2 g/m 2 plus 13 doses of ITMTX (L874B) or 4 courses of HDMTX at 4.5 g/m 2 plus 1 dose of ITMTX (I874B). Results. The 7-year probabilities (ยฑ SE) of CNS relapse-free survival were 97.3% ยฑ 2.6% (L874A, n = 41) vs. 90.3% ยฑ 5.3% (L874B, n = 39) (P = 0.25) in the LR patients, and 100% (I874A, n = 55) vs. 78.5% ยฑ 6.5% (I874B, n = 54) (P = 0.002) in the IR patients. The corresponding disease-free survival (DFS) rates were 79.4% ยฑ 6.5% vs. 74.4% ยฑ 7.3% (P = 0.62) in the LR group and 63.3% ยฑ 6.8% vs. 58.3% ยฑ 7.2% (P = 0.66) in the IR group. Thus, the HDMTX regimen could not provide better protection of CNS relapse as compared with the CRT regimen, although their overall efficacy was not significantly different. In the ALL 911 study, intensive systemic chemotherapy with extended i,t, injections of MTX plus cytarabine achieved a high CNS relapsefree survival (98% ยฑ 1.9% at 7 years) and a favorable DFS (85.5% ยฑ 5% at 7 years) in the IR patients. The patients in the high-risk (HR) group in both ALL874 and ALL911 studies received the 18-Gy or 24-Gy CRT with intensive systemic chemotherapy. Their 7-year probabilities of CNS relapse-free survival ranged from 88% to 95%, among which the T-ALL patients had a risk of CNS leukemia, which was 3-4 times higher compared with B-precursor ALL patients. Conclusions. These results indicate that long-term intrathecal CNS prophylaxis as well as appropriate systemic therapy for the non-high-risk patients can provide protection against CNS relapse equivalent to that provided by cranial irradiation.